Transcription
I personally studied about 300 guides on working with prompts for ChatGPT and selected the seven most powerful techniques. After I studied and understood these guides, I realized that 90% of people use ChatGPT incorrectly. And it's all about how to write commands correctly. And in this video is a concentrate of the best findings. And yes, they are adapted specifically for ordinary people, not for programmers. You are on the Neuro-Rhetoric channel. And let's go. Let's dive into these techniques for ChatGPT. The methods we will consider today will improve your interaction with ChatGPT literally by orders of magnitude. And yes, I'll say right away that these techniques are suitable for all neural networks. For example, they can also be successfully used in the free Deepseek. And for this video, I created this document with techniques, which we will study step by step today. I am recording this video because I have heard from other people that neural networks work very poorly. So, neural networks work well, you just need to manage them correctly. Well, let's move on to my mind map. I compiled it based on the guides I studied. I already told you about this. Plus, based on my own experience and observations. And look, first of all, I recommend understanding one problem that has recently appeared in ChatGPT. The founder of OpenAI, Sam Altman, also acknowledges this problem. And the problem is that, especially recently, ChatGPT has started to somehow excessively appease the user, try to please, excessively encourage. Well, in general, you understand. He behaves like a person who wants to be very liked, but this doesn't suit us, because we use ChatGPT for work, for our own development, and appeasement won't work for us. How to solve this problem? For this, I created this command. You are my strategic partner and advisor. Your task is to be extremely honest, direct, and so on. do not strive to please, do not smooth over the rough edges, emphasize weak points, and so on. This command will be in the guide, you will receive it, you can copy and paste it into ChatGPT. Look, you can use this in two ways. The first way, we go to customize ChatGPT and in the field, what characteristics should ChatGPT have, paste this command. You can also add it to memory. I entered this command, told him to add it to my memory, and he added it. And from that moment on, ChatGPT became my strategic partner, not someone who constantly tries to please me. I'm just tired of it already. And before we go further, I will say that I have prepared two top personal prompts for you, which you will enter into ChatGPT and develop with it, and reach a new level. These two prompts and these techniques are something truly powerful. And now we are moving on. And the next feature is to define the criteria for a quality result. ChatGPT sometimes does something that you don't like. Well, let's tell it what result you want and what you expect from it. And when you tell it what you expect from it, then it strictly stays within the framework of your expectations. You minimize the risk of it coming up with something inappropriate, too dry, or too general. Therefore, a brilliant prompt, which we are creating today and which you will also receive, must contain the criteria for the result you want. For all tasks, these criteria will be unique, and they can be created in two ways. The first way is to simply write them yourself, what you need and what you expect from it, what type of response, and so on. And the second is to give it a task, and then ask it what criteria the best solution for this task would have. It will tell you them, then you will add them to the prompt, and it will follow them precisely. As a result, you will get exactly what you expect. And let's move on to the next technique. It is quite well-known, many already know it, and that is to assign a specific role to ChatGPT. But there is one small nuance that I discovered while studying one of the scientific papers dedicated to prompts. For example, we need to develop marketing materials. And we can tell it: "You are a marketer." Okay, that's already better than nothing, but I recommend going a bit deeper. We tell it that you are a marketer in the style of Harry Weber. He is a top marketer, I like him, so I added him here. It is not necessary to insert exactly him, who writes lively, energetic texts for an audience of 25-35 years old in simple, understandable language that engages and intrigues. And this is the role that ChatGPT receives. And the role itself, which is described in detail, is precisely the start of a brilliant prompt. And, as you can see, to improve the wording, I added not only the role but also the communication style, the level of professionalism, and the target audience. And I prepared such a simple template for defining the role. You will definitely receive it. And now we are moving on. And this also relates to the role, but here something needs to be clarified. Sometimes you have a large task that has several stages. For example, in compiling the same marketing strategy, we have research, then marketing itself, and, for example, then analysis of results. I have simplified this now to show you that there are different, dissimilar stages within a marketing study. And when I personally see a large task, I want to divide it into stages and can ask ChatGPT: "What real profession or mindset is needed to best solve my specific task?" And if the task is complex and multi-stage, for example, a business development strategy, you must change the role at each stage step by step. And, for example, at the idea generation stage, you are a creative strategist. At the risk analysis stage, you are a risk manager. If you divide a large task into different roles for ChatGPT and complete it sequentially, the result of this marketing strategy will be much better. I have personally verified this through my own experience. And from this follows the next feature. Yes, sometimes you have to solve large tasks, so we assign different roles for each stage of the task. But you also need to tell ChatGPT to solve the task sequentially, step by step. In fact, there is a technical part of ChatGPT hidden here. And this part especially concerns new models that can reason. For example, the O3 model recently appeared, and it can reason. And when we tell it, "Move step by step in your reasoning process and your thoughts," it really divides the task into steps. It thinks through these steps, and then moves through each step. You can give it this command and follow its thoughts on how it performs this task. And here is a very simple principle: break down the process, make the reasoning understandable, and show the exact outcome. And dividing the task into steps helps solve one problem, because ChatGPT sometimes skips steps and gives overly generalized and unreliable answers if it is not guided. And step-by-step logic makes the answers completely transparent. And here is an example based on the same marketing strategy: develop a strategy in five stages: market research, proposition formulation, promotion channel selection, content plan, and success metrics. You can also come up with these steps yourself if you know them, but you can also ask ChatGPT to break down a large task into stages and then pass these stages to it in the prompt. And then it will cope much better. There is one small life hack on how to further improve this prompt. You can add to it: if you need more data at some step, ask me a clarifying question before you continue. If it lacks something, it will definitely clarify it with you. Look, the next technique is about making ChatGPT give even more accurate answers and your personal result to be much better. I personally love this method and constantly use it after some research or after some work. And I make it criticize itself and check its answer. And here its task is not just to give some answer, right, it also likes to be lazy, but also to evaluate the quality of its own answer, identify weak points. Tell them to me so that I can take them into account and act further. And this technique works best when you give it as a command after it has given you some answer. That is, there is no point in inserting it into a large prompt. Therefore, after you have made the main request, add, analyze your answer, find possible errors, weak points, or simplifications. And then, after it answers or immediately, you can ask it to suggest improvements. If you find inaccuracies, suggest how they can be corrected or strengthened. And here you can even specify in what style it should criticize. And here is an example. Conduct a self-check. Find at least two weak points or possible simplifications in your answer. Suggest how to strengthen each of them. Rate your degree of confidence in the final answer from one to 100. And this, by the way, is the next feature, but I inserted it here. And as a result, what does ChatGPT's self-criticism give? Answers become more accurate and precise. ChatGPT begins to think more deeply about each point, not just give the first version it finds. Well, and we save time on this, so make ChatGPT criticize itself. And here is the next way to make it do this is to ask it to state its degree of confidence in its own answers. And, for example, you can divide it on a scale from one to 10 or from one to 100, it doesn't matter. And it will give an approximate assessment. And after that, you can ask it: "Explain why you gave this particular rating," and it will give a detailed answer, and your eyes can really be opened. And here is an example. After the answer, rate your degree of confidence from one to 100, or you can use up to ten, it's not that important, and briefly explain why you gave this particular rating. If you ask ChatGPT to criticize itself or ask it to rate its confidence level in its own answers, then at this moment it can truly become your powerful strategic partner. It no longer writes lazy answers, it answers only factually, as it is in reality. And now let's look at the entire prompt. You already understand the logic of all the steps, when to insert it, what to insert at the beginning, what to insert later. Nevertheless, we will go through it. At the beginning, we give it a detailed role, where we give it the profession, expertise, experience, and for what target audience it works, and in what style it responds. Then, if necessary, if our task is large, we ask it to perform the task step by step. If individual large steps are revealed in this task, then we can also assign a separate role for one specific step. Then we ask it to conduct self-criticism and assess the degree of confidence. Here we have already added the criteria for a quality result. You define them yourself or ask it to define them. And in the end, we ask it to be extremely honest, direct, and pragmatic. But I never add this to the prompt, because I have configured the entire ChatGPT in this spirit, and it always works with me without appeasement, but gives me direct answers as a strategic advisor. All these prompts so that you can copy and paste them. And a detailed step-by-step guide on how to use these prompts is already available in my Telegram channel. The link to the Telegram channel is in the description under this video. Be sure to subscribe to it and use these powerful techniques. Well, now we will move on to two even more powerful techniques, after which my brain literally exploded when I realized that this was possible. And we will start with personal prompt number one. Look, I have personally been using ChatGPT for a long time to generate strategies, generate ideas, improve my projects, and to make its answers most suitable for me and hit the target precisely, I set it a prompt and also added this prompt to memory. Sometimes I work separately on this prompt, where I describe in detail who I am, what I do, I tell you in detail in one area or another, but lately I have been dedicating a lot of time to the field of artificial intelligence, so I work in it. Then I describe where I am, at what level, at what income level, and so on. Then I describe what is happening. Rather, this line is suitable for cases when there are some problems or something is not liked. Since I personally always want to improve my life, reach a new level, break through, yes, I am always dissatisfied with something, yes, I will be honest. And therefore, I describe what is happening with me, rather from a negative side, what I have not achieved, yes, what problems I have, what I don't like. And then I describe where all this is going. Yes, in other words, I describe my goals, my intentions, what I want to achieve, what level of expansion, what level of income, what level of knowledge and understanding. I describe all this to it in detail, yes, I have already described it once, but sometimes I adjust it when I have some new ideas. And then I tell it that when I ask it any questions or work, it should rely on this data about me and offer solutions specifically for me that will save me time and help me achieve my big goals. This personal prompt in conjunction with another prompt, where we ask it to be direct, works just excellently. I recommend you apply it today. I am sure that you will have a desire to achieve your goals, you will understand them. Or if you already have this desire, it will increase many times over, because ChatGPT will simply blow your mind with its advice. I didn't believe it. And further, to further enhance this effect, to make a more powerful breakthrough, we ask ChatGPT to be as tough as possible. Here, of course, it is important not to overdo it and some lines from here can be deleted, but in general, you can ask it to do this: "Act as my personal strategic advisor, with the following context: 'You have an IQ of 180. You are ruthlessly honest and direct. You have built several billion-dollar companies. This is to give it a role. You have deep expertise. You care about my success, but you don't tolerate excuses,' and so on. Further, we tell it that its mission is to identify critical gaps that are hindering my growth, develop concrete action plans to eliminate these gaps, push me beyond my comfort zone, point out my blind spots, provide clear frameworks and mental models. And every answer should contain the hard truth. Then specific practical steps and conclude with a direct challenge or task. When you enter this prompt after the previous prompt, when your ChatGPT is already trained, when you have worked with it, it has a lot of memory, this prompt will simply open your eyes. I have posted this prompt in my Telegram channel. Get it via the link in the description, be sure to subscribe. And right now, watch these two videos about ChatGPT to become a true expert in this topic. Use these prompts wisely and be sure to try them today. See you in the next videos. Bye everyone.